Segment PMF Comparison Sheet

The Average Can Choose the Wrong Customer

A client-reporting product has 32 trial accounts. Thirteen have sent a report, seven have returned, four have paid, and two more have proposed pilots. The aggregate looks inconclusive.

The accounts did not arrive with the same job. Freelancers tried the product after a public launch, sent one report, and rarely came back. Small agencies used it before weekly client meetings; four of the five that sent a first report returned the next week, and three accepted the standard pilot. Larger agencies had budget but required custom roles, templates, and founder-led data work before they could begin.

The broad average mixes curiosity, recurring use, and expensive sales demand. Use this sheet when that mixture is about to drive a roadmap, acquisition, or segment decision. Its job is to show where a repeatable product may exist and what still prevents the evidence from carrying that claim.

Fix the Comparison Before Reading the Results

Name the decision first. “Which segment likes the product?” is too loose. “Which segment should receive the next six weeks of recruiting and product work?” can change what the founder does.

Then declare the evidence boundary. A segment enters the comparison only when it describes a recognizable customer, buyer, recurring job, and acquisition path. “Small business” and “people who signed up” are not useful segments.

SEGMENT COMPARISON BOUNDARY
Decision this review must inform:
Product promise being compared:
Segments included and why each is distinct:
Recruiting or acquisition window:
Observation cutoff:
Account, workspace, site, or user as the unit:
Qualified-account rule:
Activation event:
Founder-assisted activation recorded separately: yes / no
Payment or commitment being compared:
Known changes in product, price, offer, or channel during the window:
Evidence gaps that limit the comparison:

Use one activation definition only when first value really is the same. If an agency activates by sending a client-ready report while a freelancer merely exports a personal draft, the segments are testing different promises. Record both honestly, but do not rank their activation rates as if the events were equivalent.

Retention follows the customer’s clock. Weekly agencies can become eligible to return next week; quarterly teams cannot. At the cutoff, keep accounts whose next cycle has not arrived out of the retention denominator. Keep founder-assisted and product-delivered value visible throughout. An account rescued by custom work may prove urgent pain while weakening the case for a repeatable product.

Build One Evidence Record per Segment

Work from account histories, not dashboard impressions. Product events, billing records, support threads, sales notes, customer language, and a log of founder labor should reconcile to the counts below.

SEGMENT EVIDENCE RECORD
Segment and recurring job:
Buyer / operator:
Why the segment is reachable by one founder:
Acquisition sources represented:
Qualified accounts approached / committed:

Accounts eligible to activate / activated:
Typical time to first value:
Activations without help / with founder help:
First broken transition among accounts that did not activate:

Activated accounts eligible to return / returned at the natural cycle:
Value cycles completed without help / with help / missed:
Payment offered / accepted at the stated price:
Renewal, expansion, contraction, churn, or referral evidence:

Customer words attached to observed behavior:
Counterevidence and strongest counterexample:
Founder sales, setup, support, and operating time per active account:
Custom work, discounts, risk, or trust obligations:
What this segment proves:
What it cannot prove yet:

Write exact counts before percentages. “Four of five eligible accounts returned” remains inspectable; “80% retention” can conceal a tiny sample and a changing denominator. When one account supplies most of the revenue or use, say so. When a percentage moves because one account became eligible, keep the account visible.

Put Only Comparable Evidence Side by Side

The matrix earns its space because the founder must see the same chain across several segments at once. Keep each cell compact and trace it back to the full segment record. Do not add the columns into a total: strong payment cannot cancel absent retention, and easy reach cannot cancel delivery that consumes the founder.

This modeled comparison does not crown the row with the most money or the highest percentage. Freelancers are reachable, but their job rarely repeats. Multi-office agencies show budget and one retained account, but the product cannot yet deliver the promise without bespoke work. Small agencies carry the most complete chain from reachable customer to recurring value, payment, and manageable service. Their activation problem is visible enough to test.

Read for a Chain, Not a Winner

For each segment, walk forward through reach, comprehension, commitment, activation, retention, payment, repeatable acquisition, and sustainable delivery. Stop at the first link the evidence cannot support. Later evidence still matters, but it cannot repair the earlier break.

Three comparisons are especially easy to misread:

• A segment that activates readily but has no recurring job may be a good audience for a one-time product, not evidence for a subscription.
• A segment that pays before operators adopt proves purchasing intent, not recurring product value.
• A segment that retains through founder rescue proves pain and perhaps a service opportunity, not yet sustainable software delivery.

Also ask whether the segments received a fair test. A founder-recruited group may receive better qualification and onboarding than public-launch signups. A newer cohort may have a better product. One segment may have seen a discount or a different promise. Preserve those differences instead of attributing every outcome to customer type.

Small samples call for narrower claims, not invented certainty. A segment with three strong accounts can justify another bounded test. It cannot support a forecast about an entire market. A segment with thin evidence should be marked “unresolved,” then given an evidence boundary; optimism is not a substitute for observation.

Make One Allocation Decision

The result should change where the next founder hour and the next product change go. Choose a primary segment, name what happens to the others, and state what new evidence could reverse the choice.

SEGMENT ALLOCATION NOTE
For the next [time or cohort boundary], we will focus on:
Recurring job and value event:

Why this segment now:
- strongest supported links:
- first weak link:
- customer behavior and language:
- payment or commitment:
- founder load and delivery boundary:

We will test:
Expected result by the next natural cycle:
Evidence that would reject this choice:

Other segments:
- hold and observe:
- serve only within an explicit paid boundary:
- stop recruiting for now:

We will not yet build, promise, or broaden:
Review date and accounts that will be eligible by then:

For the client-reporting product, the note would focus recruiting and one guided mapping experiment on small agencies. Freelancers would receive no retention project until evidence reveals a recurring job. Multi-office agencies would be held outside the product boundary unless custom setup and approval work were explicitly priced and time-boxed. The choice would be reconsidered if the next qualified small-agency cohort failed to activate, return, or pay without increasing founder rescue.

The sheet is complete when it makes one concentration decision possible while preserving the evidence that could prove it wrong. If every segment remains a priority, the comparison has described the market without allocating the founder.